How can I create a row for values that don't exist and fill the count with 0 values?

Question

In SQL Server, I\'m running a query on users age groups on data where, for some years, there are zero users per age group. For example there were users in 2013 in the \"18-21\"

Answer 1

You do this by joining with a "fake" list of age group+years:

SELECT AGE_GROUPS.YEAR, AGE_GROUPS.AGE_GROUP, COALESCE(SUM(USERS), 0) as usercount
FROM (
    SELECT YEAR, AGE_GROUP
    FROM (
        SELECT '18-21' AS AGE_GROUP
        UNION SELECT '22-25'
        UNION SELECT '25-28'
    ) AGE_GROUPS, (SELECT DISTINCT YEAR FROM USERS) YEARS
) AGE_GROUPS
LEFT JOIN USERS ON (USERS.AGE_GROUP = AGE_GROUPS.AGE_GROUP AND USERS.YEAR = AGE_GROUPS.YEAR)
WHERE AGE_GROUPS.YEAR = '2014'
GROUP BY AGE_GROUPS.YEAR, AGE_GROUPS.AGE_GROUP

You can also simplify this, assuming that your USERS table has all possible age groups ignoring a specific year:

SELECT AGE_GROUPS.YEAR, AGE_GROUPS.AGE_GROUP, COALESCE(SUM(USERS), 0) as usercount
FROM (
    SELECT YEAR, AGE_GROUP
    FROM (SELECT DISTINCT AGE_GROUP FROM USERS) AGE_GROUPS, (SELECT DISTINCT YEAR FROM USERS) YEARS
) AGE_GROUPS
LEFT JOIN USERS ON (USERS.AGE_GROUP = AGE_GROUPS.AGE_GROUP AND USERS.YEAR = AGE_GROUPS.YEAR)
WHERE AGE_GROUPS.YEAR = '2014'
GROUP BY AGE_GROUPS.YEAR, AGE_GROUPS.AGE_GROUP